5 Commits

Author	SHA1	Message	Date
saas-backend-agent	114ce95c04	fix(gpu-encoder): add pre-warm + first-byte short timeout to guard against P4000 cold-start 150s stalls ... P0 issue: 2026-09-27 staging job a15a4667 observed 149.6s delay between worker POST and P4000 starting mezzanine download (nginx access log confirmed GET mezz at 14:12:55 vs POST at 14:10:26). Root cause is server/network-side (likely Tailscale DERP hole-punching or httpx connection pool rebuild after long idle), hard to reproduce on demand. This client-side fix adds two layers of defense: 1. pre_warm: before POST, if idle >60s since last successful comm, send a GET /health with 3s timeout to warm the Tailscale path. Failure is non-fatal; POST proceeds anyway. 2. post_first_byte_timeout (default 20s): split POST timeout into two phases — first byte uses short timeout so a stalled link fails fast and caller can CPU-fallback; after first byte, relax to ffmpeg_timeout+60s for real encode+upload. Both configurable via settings.gpu_encode_pre_warm / settings.gpu_encode_post_first_byte_timeout. Also updates check_health() to refresh _last_ok_ts on healthy response.	2026-09-27 23:13:27 +08:00
xiaoxia	e4723bfb1b	perf(gpu): mezzanine relay直传跳过公网OSS(-18s), preset veryfast(-38%), 修复CI staging IP (#2063) ... Co-authored-by: xiaoxia <dev@xiaoxiajianji.com> Co-committed-by: xiaoxia <dev@xiaoxiajianji.com>	2026-09-27 14:48:54 +08:00
xiaoxia-agent	30f16c4771	fix(gpu-encoder): split relay_base_url into external/internal URLs ... - Add gpu_encode_relay_internal_base_url config (Docker DNS: http://xiaoxia-api-staging:8000) - P4000 PUT uses external URL via nginx (host:8092 plain HTTP to avoid SSL issues) - Worker GET/DELETE uses internal URL (direct Docker network, bypasses UFW/nginx) - Falls back to relay_base_url when internal URL not set (local dev) On staging: - New nginx vhost on :8092 provides plain HTTP proxy for P4000 relay PUTs - Docker network worker→API reachable at xiaoxia-api-staging:8000 (verified)	2026-09-26 20:43:21 +08:00
CI Bot	39c064ba1c	style: auto-format with black + isort + ruff + prettier [skip ci-format-check]	2026-09-26 20:43:21 +08:00
saas-backend-agent	105ab54059	feat(gpu): 接入 P4000 NVENC 硬件编码加速 ... - packages/config/base.py: 新增 GPU_ENCODE_* 配置项（开关/endpoint/relay/secret/编码参数/fallback） - packages/shared/gpu_encoder.py: 新增 GpuEncoderClient，封装 health 探测 + mezzanine 上传 OSS + P4000 nvenc 编码 + relay 回传下载，失败抛 GpuEncodeError 触发 CPU 降级 - apps/api/app/api/routes/gpu_relay.py: 新增 internal PUT/GET/DELETE /api/v1/internal/gpu-relay/{key}（token 鉴权），P4000 PUT 编码结果，worker GET 下载 - apps/api/app/api/router.py: 注册 gpu_relay_router - apps/worker/video_processing/unified_render_service.py: _execute_ffmpeg 和 _render_pass_through 尝试 GPU 路径：CPU ultrafast mezzanine → P4000 nvenc → 输出到最终路径；任何失败自动回退到原 CPU libx264 路径 - apps/worker/worker_app/tasks/_startup.py: worker_ready 时探测 P4000 健康并打日志 - tests/unit/test_gpu_encoder.py: GpuEncoderClient 单测（health/sync 调用/失败/fallback/relay URL） 架构： - 输入：CPU 输出 libx264 ultrafast mezzanine → 上传 OSS 临时前缀 → P4000 签名 URL 下载 - 输出：P4000 PUT → 宿主机 nginx（tailscale:80）→ API /api/ 反代 → gpu_relay 路由落盘到 generated/gpu_relay/ - 回传：worker 通过 docker 网络 http://xiaoxia-api-staging:8000 GET 下载最终 mp4 到 output_path - 降级：GPU 任何环节异常（health/上传/编码/回传/下载）→ 原 CPU 路径继续执行，不影响成片	2026-09-26 20:43:21 +08:00